[Seaside-dev] [seaside commit] r182 - wiki

Julian Fitzell jfitzell at gmail.com
Sun Feb 1 23:16:20 UTC 2009


Ah yes, I see what you mean. Yeah, we definitely don't have a linear
dependency set. In the long run, the load order won't matter anyway...
we just need to find time to address the lack of proper tools. I guess
we shouldn't worry about it to much for now. The Functional tests
package is unusual in that it doesn't test any other particular
package, though I sort of consider it an extra set of tests of
Environment since it depends on having basically all the base packages
loaded so in that sense it's in the right place on the wiki page.

I did update the dependency graph on that page today and removed 3
sets of circular dependencies so hopefully that will help you. Note
that the graph is only showing strictly static dependencies so it
doesn't necessarily reflect the order you need to load things in in
order for them to work, just for them to load. I also jsut filed a
couple of bugs where I saw obvious errors in the graph. I need to dig
into that code and see whether we added support for including the
WAPackage dependency info into the graph as well. Then we could
generate a full dependency graph as well as do some error checking
between the two sets.

Julian

On Sun, Feb 1, 2009 at 9:30 PM, James Foster <Smalltalk at jgfoster.net> wrote:
> Julian,
>
> Thanks for keeping an eye on things here. Actually I'm a bit confused myself
> (and I'm sure it shows). Actually, I suspect that we are in a situation in
> which some of the Traditional packages are not dependent on each other, so
> any ordering of them in a linear list can be misleading.
>
> Here is my brief reconstruction of things:
>
> 1. At one time some of the tests were in a separate group on the LoadOrder
> Wiki with the note that they could be loaded after their package was loaded.
> In revision 171 I refactored the LoadOrder page to put the tests after the
> package they test (including Seaside-Tests-Flow after Seaside-Flow), but
> intended the implication that tests were optional. This seemed to work well
> through the Basic group. Once I got to the Traditional group things started
> falling apart.
>
> 2. Soon, I found that Seaside-Tests-Flow depend on Seaside-Tests-Functional
> (#326), contrary to the previous note that tests could be loaded after their
> package. Based on comment #1 in Issue #326
> (http://lists.squeakfoundation.org/pipermail/seaside-dev/2009-January/002658.html),
> I made revision 179 that put Seaside-Flow before Seaside-Tests-Functional.
>
> 3. Based on
> http://lists.squeakfoundation.org/pipermail/seaside-dev/2009-January/002667.html I
> had the impression that Seaside-Continuation should come before
> Seaside-Flow, so I made revision 180 to reflect that
> (http://lists.squeakfoundation.org/pipermail/seaside-dev/2009-January/002672.html).
>
> I think that because of #2 above I wanted Seaside-Tests-Functional to come
> before Seaside-Tests-Flow. Putting them after Seaside-Flow gives the
> impression that the functional tests require Flow, and I don't think that is
> the case. I think we are in a situation in which the ordering of the
> functional tests and Continuations could be done either way, as long as they
> are both loaded before Seaside-Tests-Flow. It seems that either ordering
> implies that the other would not work, and thus raises questions. I suspect
> that more of these will arise as we work through the Traditional list. Does
> Seaside-Examples require Seaside-Development and Seaside-WebTools? Does
> Seaside-Development require Seaside-Squeak-Continuations?
>
> I'd welcome clarification/help here. If you want to reorder the list to what
> it "should" be then I can start over "testing" it. Would that make sense?
>
> James
>
>
> On Feb 1, 2009, at 11:27 AM, Julian Fitzell wrote:
>
>> James,
>>
>> I'm having trouble following all of this indirectly, but I don't see
>> why Continuation packages would depend on the functional tests. Are
>> you finding that they do?
>>
>> Julian
>>
>> On Sun, Feb 1, 2009 at 6:15 PM,  <codesite-noreply at google.com> wrote:
>>>
>>> Author: WeybridgeWay
>>> Date: Sun Feb  1 09:05:29 2009
>>> New Revision: 182
>>>
>>> Modified:
>>> wiki/LoadOrder.wiki
>>>
>>> Log:
>>> Continue moving functional tests earlier. Remove ANSI tests. --James
>>>
>>> Modified: wiki/LoadOrder.wiki
>>>
>>> ==============================================================================
>>> --- wiki/LoadOrder.wiki (original)
>>> +++ wiki/LoadOrder.wiki Sun Feb  1 09:05:29 2009
>>> @@ -5,7 +5,6 @@
>>>
>>> = Basic =
>>> * Seaside-Platform
>>> -  * Seaside-Tests-ANSI
>>> * Seaside-Squeak-Platform
>>> * Seaside-Tests-Platform
>>> * Seaside-Tests-Squeak-Platform
>>> @@ -26,12 +25,13 @@
>>> * Seaside-Tests-Environment (for defects see Issue 319)
>>>
>>> = Traditional =
>>> + * (functional tests)
>>> +  * Seaside-Tests-Functional
>>> +  * Seaside-Tests-Squeak-Functional
>>> * Seaside-Squeak-Continuation
>>> * Seaside-Squeak-Continuation
>>> * Seaside-Flow
>>> * Seaside-Squeak-Flow
>>> -  * Seaside-Tests-Functional
>>> -  * Seaside-Tests-Squeak-Functional
>>> * Seaside-Tests-Flow
>>> * Seaside-!WebTools
>>> * Seaside-Squeak-!WebTools
>>> _______________________________________________
>>> seaside-dev mailing list
>>> seaside-dev at lists.squeakfoundation.org
>>> http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev
>>>
>> _______________________________________________
>> seaside-dev mailing list
>> seaside-dev at lists.squeakfoundation.org
>> http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev
>>
>
> _______________________________________________
> seaside-dev mailing list
> seaside-dev at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev
>


More information about the seaside-dev mailing list